Page 293 - Fundamentals of Light Microscopy and Electronic Imaging

276 DIGITAL CCD MICROSCOPY

Signal-to-Noise (S/N) Ratio

S/N ratio is used to describe the photometric accuracy of an object’s signal. In qualitative
terms, we use S/N ratios to describe the clarity and visibility of objects in an image. S/N
ratio is calculated as the object signal (total signal minus contributing background signal)
divided by the noise of the surrounding background (standard deviation of the back-
ground signal). When used to describe the imaging performance of a CCD camera, S/N
is calculated in terms of a single pixel, and the calculation is always based on the number
of electrons comprising the signal and the noise. The importance of S/N is easily appre-
ciated when we examine a dim, grainy image, where the amplitude of the object signal is
small and the read noise of the camera is a principal noise component in the image. In a
side-by-side comparison of the imaging performance of two cameras with read noises
differing by a factor of 2, the difference is clear: The camera with the lower read noise
produces the clearer image. The effect of high camera noise can also be observed on the
computer monitor or on a print in half-saturated, moderately exposed images. S/N char-
acteristics are even more significant for those using a CCD camera as a photometer to
monitor changes in light intensity, such as in fluorescence experiments involving FRAP,
FRET, or ratio imaging of fluorescent dyes. In this case, plots of light intensity over time
are smoother and more accurate when image sequences exhibit high S/N values. We will
examine S/N theory and its applications in greater detail in Chapter 15.

BENEFITS OF DIGITAL CCD CAMERAS

Low-light sensitivity Comparable film ASA 100,000 for equivalent S/N ratio;
useful range of light intensity, 4–5 orders of magnitude; 3–4
orders of magnitude more sensitive than video or film
Low instrument noise Cooling and electronics give high S/N and clear visible

images; read noise/pixel as low as 3–5e /pixel for biologi-
cal cameras
Spatial resolution Small (4–9 m) pixels preserve optical resolution even at
low magnifications
Time resolution up to 10 frame/s for full-frame megapixel chip
Dynamic range Thousands of gray levels; 10–16 bit vs. 6–8 bit for video
Less noise Up to several hundred times less than video
Digital output Pixels give quantitative value of light intensity; for 12 bit
digitizer, up to 4096 gray levels
Linear response 0.1% nonlinearity over 4 orders of magnitude
Flexible readout Subarray, binning modes allow optimization of space, time,
intensity

REQUIREMENTS AND DEMANDS OF DIGITAL CCD IMAGING

High costs $10,000–$20,000 for a high-performance system
Personnel Requires experienced user/computer technician
Additional components Electromechanical shutter, filter wheel, computer interface
card, RAM upgrade, acquisition/processing software